seq_runtime/
serialize.rs

1//! Serialization of Seq Values
2//!
3//! This module provides a serializable representation of Seq runtime values.
4//! It enables Value persistence and exchange with external systems.
5//!
6//! # Use Cases
7//!
8//! - **Actor persistence**: Event sourcing and state snapshots
9//! - **Data pipelines**: Arrow/Parquet integration
10//! - **IPC**: Message passing between processes
11//! - **Storage**: Database and file persistence
12//!
13//! # Why TypedValue?
14//!
15//! The runtime `Value` type contains arena-allocated strings (`SeqString`)
16//! which aren't directly serializable. `TypedValue` uses owned `String`s
17//! and can be serialized with serde/bincode.
18//!
19//! # Why BTreeMap instead of HashMap?
20//!
21//! `TypedValue::Map` uses `BTreeMap` (not `HashMap`) for deterministic serialization.
22//! This ensures that the same logical map always serializes to identical bytes,
23//! which is important for:
24//! - Content-addressable storage (hashing serialized data)
25//! - Reproducible snapshots for testing and debugging
26//! - Consistent behavior across runs
27//!
28//! The O(n log n) insertion overhead is acceptable since serialization is
29//! typically infrequent (snapshots, persistence) rather than on the hot path.
30//!
31//! # Performance
32//!
33//! Uses bincode for fast, compact binary serialization.
34//! For debugging, use `TypedValue::to_debug_string()`.
35//!
36//! # Byte-cleanliness boundary
37//!
38//! `TypedValue::String` and `TypedMapKey::String` hold owned `String` —
39//! UTF-8 by definition. Conversion from a runtime `Value::String`
40//! (which is byte-clean and may carry arbitrary bytes) goes through
41//! `as_str_or_empty()`: invalid UTF-8 collapses to the empty string.
42//! That is the deliberate, narrow contract of this module — it serves
43//! the *text-shaped* payloads of actor persistence, IPC, and
44//! Arrow/Parquet pipelines, not arbitrary binary blobs.
45//!
46//! Programs that need to persist binary `String` payloads should
47//! base64- or hex-encode them at the Seq layer before handing them
48//! to `serialize`, or use a binary-aware transport (file slurp/spit,
49//! HTTP body, channel send) which retains bytes verbatim.
50
51use crate::seqstring::global_string;
52use crate::value::{MapKey as RuntimeMapKey, Value, VariantData};
53use serde::{Deserialize, Serialize};
54use std::collections::{BTreeMap, HashMap};
55use std::sync::Arc;
56
57/// Error during serialization/deserialization
58#[derive(Debug)]
59pub enum SerializeError {
60    /// Cannot serialize quotations (code)
61    QuotationNotSerializable,
62    /// Cannot serialize closures
63    ClosureNotSerializable,
64    /// Cannot serialize channels (runtime state)
65    ChannelNotSerializable,
66    /// Bincode encoding/decoding error (preserves original error for debugging)
67    BincodeError(Box<bincode::Error>),
68    /// Invalid data structure
69    InvalidData(String),
70    /// Non-finite float (NaN or Infinity)
71    NonFiniteFloat(f64),
72}
73
74impl std::fmt::Display for SerializeError {
75    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
76        match self {
77            SerializeError::QuotationNotSerializable => {
78                write!(f, "Quotations cannot be serialized - code is not data")
79            }
80            SerializeError::ClosureNotSerializable => {
81                write!(f, "Closures cannot be serialized - code is not data")
82            }
83            SerializeError::ChannelNotSerializable => {
84                write!(f, "Channels cannot be serialized - runtime state")
85            }
86            SerializeError::BincodeError(e) => write!(f, "Bincode error: {}", e),
87            SerializeError::InvalidData(msg) => write!(f, "Invalid data: {}", msg),
88            SerializeError::NonFiniteFloat(v) => {
89                write!(f, "Cannot serialize non-finite float: {}", v)
90            }
91        }
92    }
93}
94
95impl std::error::Error for SerializeError {
96    fn source(&self) -> Option<&(dyn std::error::Error + 'static)> {
97        match self {
98            SerializeError::BincodeError(e) => Some(e.as_ref()),
99            _ => None,
100        }
101    }
102}
103
104impl From<bincode::Error> for SerializeError {
105    fn from(e: bincode::Error) -> Self {
106        SerializeError::BincodeError(Box::new(e))
107    }
108}
109
110/// Serializable map key types
111///
112/// Subset of TypedValue that can be used as map keys.
113/// Mirrors runtime `MapKey` but with owned strings.
114#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq, PartialOrd, Ord, Hash)]
115pub enum TypedMapKey {
116    Int(i64),
117    Bool(bool),
118    String(String),
119}
120
121impl TypedMapKey {
122    /// Convert to a TypedValue
123    pub fn to_typed_value(&self) -> TypedValue {
124        match self {
125            TypedMapKey::Int(v) => TypedValue::Int(*v),
126            TypedMapKey::Bool(v) => TypedValue::Bool(*v),
127            TypedMapKey::String(v) => TypedValue::String(v.clone()),
128        }
129    }
130
131    /// Convert from runtime MapKey
132    pub fn from_runtime(key: &RuntimeMapKey) -> Self {
133        match key {
134            RuntimeMapKey::Int(v) => TypedMapKey::Int(*v),
135            RuntimeMapKey::Bool(v) => TypedMapKey::Bool(*v),
136            RuntimeMapKey::String(s) => TypedMapKey::String(s.as_str_or_empty().to_string()),
137        }
138    }
139
140    /// Convert to runtime MapKey (requires global string allocation)
141    pub fn to_runtime(&self) -> RuntimeMapKey {
142        match self {
143            TypedMapKey::Int(v) => RuntimeMapKey::Int(*v),
144            TypedMapKey::Bool(v) => RuntimeMapKey::Bool(*v),
145            TypedMapKey::String(s) => RuntimeMapKey::String(global_string(s.clone())),
146        }
147    }
148}
149
150/// Serializable representation of Seq Values
151///
152/// This type mirrors `Value` but uses owned data suitable for serialization.
153/// Quotations and closures cannot be serialized (they contain code, not data).
154#[derive(Debug, Clone, Serialize, Deserialize, PartialEq)]
155pub enum TypedValue {
156    Int(i64),
157    Float(f64),
158    Bool(bool),
159    String(String),
160    /// Symbol (interned identifier)
161    Symbol(String),
162    /// Map with typed keys and values
163    Map(BTreeMap<TypedMapKey, TypedValue>),
164    /// Variant with tag (symbol name) and fields
165    Variant {
166        tag: String,
167        fields: Vec<TypedValue>,
168    },
169}
170
171impl TypedValue {
172    /// Convert from runtime Value
173    ///
174    /// Returns error if Value contains:
175    /// - Code (Quotation/Closure) - not serializable
176    /// - Non-finite floats (NaN/Infinity) - could cause logic issues
177    pub fn from_value(value: &Value) -> Result<Self, SerializeError> {
178        match value {
179            Value::Int(v) => Ok(TypedValue::Int(*v)),
180            Value::Float(v) => {
181                if !v.is_finite() {
182                    return Err(SerializeError::NonFiniteFloat(*v));
183                }
184                Ok(TypedValue::Float(*v))
185            }
186            Value::Bool(v) => Ok(TypedValue::Bool(*v)),
187            Value::String(s) => Ok(TypedValue::String(s.as_str_or_empty().to_string())),
188            Value::Symbol(s) => Ok(TypedValue::Symbol(s.as_str_or_empty().to_string())),
189            Value::Map(map) => {
190                let mut typed_map = BTreeMap::new();
191                for (k, v) in map.iter() {
192                    let typed_key = TypedMapKey::from_runtime(k);
193                    let typed_value = TypedValue::from_value(v)?;
194                    typed_map.insert(typed_key, typed_value);
195                }
196                Ok(TypedValue::Map(typed_map))
197            }
198            Value::Variant(data) => {
199                let mut typed_fields = Vec::with_capacity(data.fields.len());
200                for field in data.fields.iter() {
201                    typed_fields.push(TypedValue::from_value(field)?);
202                }
203                Ok(TypedValue::Variant {
204                    tag: data.tag.as_str_or_empty().to_string(),
205                    fields: typed_fields,
206                })
207            }
208            Value::Quotation { .. } => Err(SerializeError::QuotationNotSerializable),
209            Value::Closure { .. } => Err(SerializeError::ClosureNotSerializable),
210            Value::Channel(_) => Err(SerializeError::ChannelNotSerializable),
211            Value::WeaveCtx { .. } => Err(SerializeError::ChannelNotSerializable), // Weaves contain channels
212        }
213    }
214
215    /// Convert to runtime Value
216    ///
217    /// Note: Strings are allocated as global strings (not arena)
218    /// to ensure they outlive any strand context.
219    pub fn to_value(&self) -> Value {
220        match self {
221            TypedValue::Int(v) => Value::Int(*v),
222            TypedValue::Float(v) => Value::Float(*v),
223            TypedValue::Bool(v) => Value::Bool(*v),
224            TypedValue::String(s) => Value::String(global_string(s.clone())),
225            TypedValue::Symbol(s) => Value::Symbol(global_string(s.clone())),
226            TypedValue::Map(map) => {
227                let mut runtime_map = HashMap::new();
228                for (k, v) in map.iter() {
229                    runtime_map.insert(k.to_runtime(), v.to_value());
230                }
231                Value::Map(Box::new(runtime_map))
232            }
233            TypedValue::Variant { tag, fields } => {
234                let runtime_fields: Vec<Value> = fields.iter().map(|f| f.to_value()).collect();
235                Value::Variant(Arc::new(VariantData::new(
236                    global_string(tag.clone()),
237                    runtime_fields,
238                )))
239            }
240        }
241    }
242
243    /// Try to convert to a map key (fails for Float, Map, Variant)
244    pub fn to_map_key(&self) -> Result<TypedMapKey, SerializeError> {
245        match self {
246            TypedValue::Int(v) => Ok(TypedMapKey::Int(*v)),
247            TypedValue::Bool(v) => Ok(TypedMapKey::Bool(*v)),
248            TypedValue::String(v) => Ok(TypedMapKey::String(v.clone())),
249            TypedValue::Float(_) => Err(SerializeError::InvalidData(
250                "Float cannot be a map key".to_string(),
251            )),
252            TypedValue::Map(_) => Err(SerializeError::InvalidData(
253                "Map cannot be a map key".to_string(),
254            )),
255            TypedValue::Variant { .. } => Err(SerializeError::InvalidData(
256                "Variant cannot be a map key".to_string(),
257            )),
258            TypedValue::Symbol(v) => Ok(TypedMapKey::String(v.clone())),
259        }
260    }
261
262    /// Serialize to binary format (bincode)
263    pub fn to_bytes(&self) -> Result<Vec<u8>, SerializeError> {
264        bincode::serialize(self).map_err(SerializeError::from)
265    }
266
267    /// Deserialize from binary format (bincode)
268    pub fn from_bytes(bytes: &[u8]) -> Result<Self, SerializeError> {
269        bincode::deserialize(bytes).map_err(SerializeError::from)
270    }
271
272    /// Convert to human-readable debug string
273    pub fn to_debug_string(&self) -> String {
274        match self {
275            TypedValue::Int(v) => format!("{}", v),
276            TypedValue::Float(v) => format!("{}", v),
277            TypedValue::Bool(v) => format!("{}", v),
278            TypedValue::String(v) => format!("{:?}", v),
279            TypedValue::Symbol(v) => format!(":{}", v),
280            TypedValue::Map(m) => {
281                let entries: Vec<String> = m
282                    .iter()
283                    .map(|(k, v)| format!("{}: {}", key_to_debug_string(k), v.to_debug_string()))
284                    .collect();
285                format!("{{ {} }}", entries.join(", "))
286            }
287            TypedValue::Variant { tag, fields } => {
288                if fields.is_empty() {
289                    format!("(Variant#{})", tag)
290                } else {
291                    let field_strs: Vec<String> =
292                        fields.iter().map(|f| f.to_debug_string()).collect();
293                    format!("(Variant#{} {})", tag, field_strs.join(" "))
294                }
295            }
296        }
297    }
298}
299
300fn key_to_debug_string(key: &TypedMapKey) -> String {
301    match key {
302        TypedMapKey::Int(v) => format!("{}", v),
303        TypedMapKey::Bool(v) => format!("{}", v),
304        TypedMapKey::String(v) => format!("{:?}", v),
305    }
306}
307
308/// Extension trait for Value to add serialization methods
309pub trait ValueSerialize {
310    /// Convert to serializable TypedValue
311    fn to_typed(&self) -> Result<TypedValue, SerializeError>;
312
313    /// Serialize directly to bytes
314    fn to_bytes(&self) -> Result<Vec<u8>, SerializeError>;
315}
316
317impl ValueSerialize for Value {
318    fn to_typed(&self) -> Result<TypedValue, SerializeError> {
319        TypedValue::from_value(self)
320    }
321
322    fn to_bytes(&self) -> Result<Vec<u8>, SerializeError> {
323        TypedValue::from_value(self)?.to_bytes()
324    }
325}
326
327#[cfg(test)]
328mod tests;
seq_runtime/serialize.rs

seq_runtime/
serialize.rs
